Bug 2005759 - Managed Clusters show "Resuming" status after upgrading from ACM 2.3.2 to ACM 2.4.0
Summary: Managed Clusters show "Resuming" status after upgrading from ACM 2.3.2 to ACM...
Keywords:
Status: CLOSED WORKSFORME
Alias: None
Product: Red Hat Advanced Cluster Management for Kubernetes
Classification: Red Hat
Component: Console
Version: rhacm-2.3.z
Hardware: Unspecified
OS: Unspecified
medium
low
Target Milestone: ---
: rhacm-2.5
Assignee: Yuanyuan He
QA Contact: Eveline Cai
Christopher Dawson
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2021-09-20 05:31 UTC by Utkarsh Wagh
Modified: 2025-04-04 13:13 UTC (History)
11 users (show)

Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
Environment:
Last Closed: 2022-01-12 08:03:56 UTC
Target Upstream Version:
Embargoed:
ming: rhacm-2.4+
bot-tracker-sync: needinfo+


Attachments (Terms of Use)
Screenshot of the ACM console reflecting message ***Upgrading to 4.7.28***. (73.78 KB, image/png)
2021-09-20 05:31 UTC, Utkarsh Wagh
no flags Details


Links
System ID Private Priority Status Summary Last Updated
Github open-cluster-management backlog issues 16298 0 None None None 2021-09-20 12:52:35 UTC

Description Utkarsh Wagh 2021-09-20 05:31:16 UTC
Created attachment 1824486 [details]
Screenshot of the ACM console reflecting message ***Upgrading to 4.7.28***.

Description of problem: After upgrading the RHOCP cluster from v4.7.28 to v4.7.29, the upgrade is successful and the overall cluster is healthy and stable.
The output of ***$ oc get clusterversion*** is reflecting OCP v4.7.29, However, the ACM console is reflecting it as ***Upgrading to 4.7.28***.
~~~
$ oc get clusterversion

[[ us-acm ]]
NAME      VERSION   AVAILABLE   PROGRESSING   SINCE   STATUS
version   4.7.29    True        False         31h     Cluster version is 4.7.29

[[ us-test ]]
NAME      VERSION   AVAILABLE   PROGRESSING   SINCE   STATUS
version   4.7.29    True        False         31h     Cluster version is 4.7.29

[[ us-dev ]]
NAME      VERSION   AVAILABLE   PROGRESSING   SINCE   STATUS
version   4.7.29    True        False         31h     Cluster version is 4.7.29

[[ us-prod ]]
NAME      VERSION   AVAILABLE   PROGRESSING   SINCE   STATUS
version   4.7.29    True        False         31h     Cluster version is 4.7.29

[[ eu-dev ]]
NAME      VERSION   AVAILABLE   PROGRESSING   SINCE   STATUS
version   4.7.29    True        False         31h     Cluster version is 4.7.29

[[ eu-prod ]]
NAME      VERSION   AVAILABLE   PROGRESSING   SINCE   STATUS
version   4.7.29    True        False         30h     Cluster version is 4.7.29
~~~
 
[1] have analyzed the must-gather report, no signs of errors caused during the upgrade.

[2] have analyzed the ns/open-cluster-management(ACM instance), no signs of errors, and all pods are up and running.

[3] have tried workaround as clearing web browser cache/cookies, accessing URL via Incognito/private window. 



Version-Release number of selected component (if applicable): RHACM v2.3.2


How reproducible:


Steps to Reproduce:
1.
2.
3.

Actual results:


Expected results: The cluster distribution version as seen in the ACM console should reflect the updated cluster version.


Additional info: The customer has upgraded six RHOCP clusters from v4.6.28 to v4.6.29, all clusters have upgraded successfully and are stable, however, three of the six clusters are still refecting ***Upgrading to 4.7.28*** in the ACM console.

I will attach the must-gather report if needed.

Comment 3 Mike Ng 2021-10-04 20:47:01 UTC
G2Bsync 933626538 comment 
 jamestalton Mon, 04 Oct 2021 16:00:33 UTC 
 G2Bsync
``` 
isUpgradeCuration =
            clusterCurator?.spec?.desiredCuration === 'upgrade' ||
            checkCuratorLatestOperation(CuratorCondition.upgrade, curatorConditions) ||
            checkCuratorLatestFailedOperation(CuratorCondition.upgrade, curatorConditions)

curatorIsUpgrading =
            isUpgradeCuration &&
            clusterCurator?.spec?.upgrade?.desiredUpdate !==
                managedClusterInfo?.status?.distributionInfo?.ocp?.version

isUpgrading =
  curatorIsUpgrading ||
  desiredVersion !== managedClusterInfo?.status?.distributionInfo?.ocp?.version
```

We need the ClusterCurator and ManagedClusterInfo resources for a cluster in this state to determine why it is displaying the wrong status.

Comment 6 Kevin Cormier 2021-11-09 18:53:56 UTC
Utkarsh, we are not able to access the support case you created. (Neither jtalton nor kcormier)

Do you know how we can access?

Comment 8 Kevin Cormier 2021-11-10 18:30:44 UTC
Yes, we are able to see these files and are investigating. Thank you!

Comment 9 Kevin Cormier 2021-11-10 19:31:54 UTC
Can you please clarify the new query from the customer: "We are unable to Select a channel from the ACM clusters page for the 4.7.30"

1. From the lastest screen captures, all of the clusters are on OCP 4.8.11 while "us-dev" shows as upgrading to 4.7.30. Is it this one that is having the problem with selecting a channel?
2. Is the option to change the channel available? Is the list of available channels not as expected? Or is there an error when you select a new channel? Or does ACM appear to allow you to change the channel, but it has no effect on the actual cluster?

Comment 10 Utkarsh Wagh 2021-11-15 16:05:51 UTC
Hello Team,

To avoid this confusion, Let me summarize things from the start.

[1] The case was initially opened for customer-facing issues as, the customer have upgraded the ***six*** clusters from v4.7.28 to v4.7.29, out of which ***three*** clusters were reflecting ***upgrading to v4.7.28*** though the upgrade was completed successfully and all the operators were on v4.7.29.

Can be clearly seen from the attachment[1] on this bug on ***2021-09-20***.

[2] Then the customer aging upgraded these six clusters from ***v4.7.29*** to ***v4.7.30***, the issue still persisted as the affected three clusters were reflecting the same ***upgrading to v4.7.28***, though the cluster was upgraded successfully and all operators were at v4.7.30.

[3] Then the customer aging upgraded the six clusters from ***v4.7.30*** to ***v4.8.11***, the issue still persisted as the affected three clusters were reflecting the same ***upgrading to ***v4.7.28***, though the cluster was upgraded successfully and all operators were at ***v4.8.11.

But this time cu had new concerns as that customer has updated some OCP clusters to ***v4.8.11*** and the current version for ACM is ***2.3.2*** and some clusters stuck in still in ***v4.7.30***.
Can be seen clearly from the output shared on the case on comment 20 on the case.
~~~
Just an observation, we are unable to Select channels from the ACM clusters page for the 4.7.30 clusters and those stuck at "Upgrading".

Name          | Distribution version                   | ...
us-dev        | OpenShift 4.7.30                       | Edit labels,                 Search cluster, Detach cluster, Destroy cluster
us-prod       | OpenShift 4.7.30 - Upgrading to 4.7.28 | Edit labels,                 Search cluster, Detach cluster, Destroy cluster
eu-prod       | OpenShift 4.7.30 - Upgrading to 4.7.28 | Edit labels,                 Search cluster, Detach cluster, Destroy cluster
eu-dev        | OpenShift 4.8.11 - Upgrading to 4.7.28 | Edit labels,                 Search cluster, Detach cluster, Destroy cluster
us-test       | OpenShift 4.8.11                       | Edit labels, Select channel, Search cluster, Detach cluster, Destroy cluster
local-cluster | OpenShift 4.8.11                       | Edit labels, Select channel, Search cluster, Detach cluster
~~~

[4] After this engineering team requested for ***ClusterCurator and ManagedClusterInfo resources for a cluster in this state to determine why it is displaying the wrong status.*** as attached to the bug on comments 4 and 7.

[5] After this, the CU has not upgraded the cluster from v4.8.11 to further.

I hope the above summarized the issue till now and clears CU's new query ***We are unable to Select a channel from the ACM clusters page for the 4.7.30***.

If need I can ask the customer for a fresh screenshot of the console, to check which clusters are still reflecting the issue ***upgrading to v4.7.28*** and which clusters are reflecting ***unable to select channel issue***.

Do let me know your thoughts on the same.

Regards,

Comment 11 Kevin Cormier 2021-12-02 20:27:26 UTC
B2Gsync We are still investigating the problem with clusters stuck showing an upgrading status.

As a workaround, you can delete the ClusterCurator resource in the cluster namespace for each cluster showing the stuck status. For example, for eu-dev, on the ACM hub cluster run:

oc -n eu-dev delete clustercurator eu-dev

For clusters that are not offering the "Select channel" option, we would need to see the ManagedClusterInfo from the hub, and also the ClusterVersion resource from the managed cluster, current as of when the UI issue is occurring. In the table above, 3 of the clusters (us-dev, us-prod, eu-prod) show version 4.7.30, but in the data you collected, they are all at 4.8.11.

In order to show the "Select channel" option, the ClusterVersion resource on the managed cluster needs to provide available channels. This is information is then reflected in the ManagedClusterInfo. If the cluster is in a state where its current version does not belong to its current selected channel, the list of available channels is not provided, so we will not have the "Select channel" option available in ACM. It can be changed by manually typing in a new channel name in the OCP console directly on the managed cluster. 

If this is the case for those clusters, in OCP under Administration > Cluster Settings, you would see an error alert saying "Version x.y.z not found in channel channel-x.y. To request update recommendations, configure a channel that supports your version."

Comment 12 Utkarsh Wagh 2021-12-03 15:42:27 UTC
Hello Team,

Thanks for your response and suggesting the workaround as above.

I have the latest update from the customer side as CU has updated the ACM version from ***2.3.3 to 2.4.0*** and now facing the issue as the clusters are stuck at the "Resuming" state and the misleading version issue is resolved ***Upgrading to 4.7.28*** from the web-console.

To elaborate more:

[1] eu-dev is on OCP v4.8.17 and reflecting error on console as "Resuming"


[2] eu-prod is on OCP v4.8.17 and reflecting error on console as "Resuming"


[3] local-cluster is on OCP v4.8.17 and reflects no error and even upgrade path is available ------> no issue for this cluster.


[4] us-dev is on OCP v4.8.17 and reflecting error on console as "Resuming"


[5] us-prod is on OCP v4.8.17 and reflecting error on console as "Resuming"


[6] us-test is on OCP v4.8.17 and reflecting error on console as "Resuming"

I will be attaching the screenshot of the console as well for your reference.

Now Cu has concerns as:

~~~
[a] Everything is OK with local-cluster, status Ready and Distribution version OpenShift 4.8.17 with Upgrade available are the expected values.

[b] The other clusters are supposed to be showing the same Ready and 4.8.17 values as local-cluster but as you can see their Status is incorrect at Resuming, their Distribution version is correct but we are not seeing the expected Upgrade available link.
~~~

Now should I suggest the workaround mentioned in your update or there is a change in action? Let me know if more data is required from the Cu side.

Regards,
Utkarsh Wagh

Comment 14 Kevin Cormier 2021-12-03 17:03:16 UTC
B2Gsync

Hi Utkarsh,

No, the workaround of deleting the ClusterCurator resources is only applicable if the UI is showing an upgrade in progress when that is not the case.

Since the Status of these clusters is "Resuming", the available updates will not appear in the UI. They are only offered if the cluster has "Ready" status.

There are 3 questions I have:

[1] Does the UI status match what is shown in the ClusterDeployment resources? (If so, we should transfer this to the server foundation squad for investigation.)

[2] Did the customer hibernate these clusters and then resume them, or is the "Resuming" status unexpected?

[3] Are the clusters with "Resuming" status actually still resuming, or can the customer access them? 

Thanks

Comment 15 Utkarsh Wagh 2021-12-08 08:50:10 UTC
Hello Team,

Following are the answers to the above questions.

[1] Does the UI status match what is shown in the ClusterDeployment resources?

-> Yes, the UI and ClusterDeployment resource status match.


[us-acm@aws-us-bastion ~]$ oc get clusterdeployment -A
NAMESPACE   NAME      PLATFORM   REGION      CLUSTERTYPE   INSTALLED   INFRAID         VERSION   POWERSTATE   AGE
eu-dev      eu-dev    aws        eu-west-1                 true        eu-dev-wtw7x    4.7.19    Resuming     159d
eu-prod     eu-prod   aws        eu-west-1                 true        eu-prod-hbpkw   4.7.19    Resuming     154d
us-dev      us-dev    aws        us-east-1                 true        us-dev-chxj5    4.7.19    Resuming     181d
us-prod     us-prod   aws        us-east-1                 true        us-prod-w2d54   4.7.19    Resuming     175d
us-test     us-test   aws        us-east-1                 true        us-test-kmrwc   4.7.19    Resuming     194d
[us-acm@aws-us-bastion ~]$


[2] Have you hibernated these clusters before and then resume them, or is the "Resuming" status unexpected?

-> We have not hibernated any of our clusters before, "Resuming" status is unexpected.

[3] Are the clusters with "Resuming" status actually still resuming, or can you be able to access them?

-> No clusters are resuming, I can access all of them.


Let me know if more information is required from Cu's side.

Regards,
Utkarsh Wagh

Comment 16 bot-tracker-sync 2021-12-08 16:22:08 UTC
G2Bsync 988891350 comment 
 KevinFCormier Wed, 08 Dec 2021 15:01:32 UTC 
 G2Bsync Utkarsh, can you please provide the full ClusterDeployment YAML?

Comment 17 daliu 2021-12-09 03:27:00 UTC
Utkarsh, please help to provide the full Clusterdeployment Yaml in eu-dev namespace and the hive-controllers pod log in hive namespace.

Comment 21 daliu 2021-12-16 10:49:00 UTC
@pmagotra Could you help confirm with customer follow https://bugzilla.redhat.com/show_bug.cgi?id=2005759#c20, if they change the certificates? or other action will change the certificates? 
If the secret `eu-dev-0-zpcgh-admin-kubeconfig` in eu-dev ns can connect to the cluster?

Comment 22 Utkarsh Wagh 2021-12-22 09:09:35 UTC
Hello Team,

There is an update from the customer for the following:

~~~
[1] If you had changed the certificates?
[2] or perform other actions that will change the certificates?
[3] Can you check If the secret `eu-dev-0-zpcgh-admin-kubeconfig` in eu-dev ns can connect to the cluster?
~~~

According to the customer:

[1] He followed Adding API server certificates to add our corporate-signed API certificates as part of standard security configurations on all clusters, below is a sample configuration on eu-dev cluster.

~~~
[eu-dev@aws-us-bastion ~]$ oc describe secret -n openshift-config bi-api-certificate
Name:         bi-api-certificate
Namespace:    openshift-config
Labels:       <none>
Annotations:  <none>

Type:  kubernetes.io/tls

Data
====
tls.crt:  8422 bytes
tls.key:  1704 bytes
[eu-dev@aws-us-bastion ~]$ oc describe apiserver
Name:         cluster
Namespace:
Labels:       <none>
Annotations:  include.release.openshift.io/self-managed-high-availability: true
              include.release.openshift.io/single-node-developer: true
              oauth-apiserver.openshift.io/secure-token-storage: true
              release.openshift.io/create-only: true
API Version:  config.openshift.io/v1
Kind:         APIServer
Metadata:
  Creation Timestamp:  2021-07-01T09:07:49Z
  Generation:          4
  Managed Fields:
    API Version:  config.openshift.io/v1
    Fields Type:  FieldsV1
    fieldsV1:
      f:metadata:
        f:annotations:
          .:
          f:include.release.openshift.io/self-managed-high-availability:
          f:include.release.openshift.io/single-node-developer:
          f:oauth-apiserver.openshift.io/secure-token-storage:
          f:release.openshift.io/create-only:
      f:spec:
        .:
        f:audit:
          .:
          f:profile:
    Manager:      cluster-version-operator
    Operation:    Update
    Time:         2021-07-01T09:07:49Z
    API Version:  config.openshift.io/v1
    Fields Type:  FieldsV1
    fieldsV1:
      f:spec:
        f:additionalCORSAllowedOrigins:
    Manager:      OpenAPI-Generator
    Operation:    Update
    Time:         2021-07-14T09:50:10Z
    API Version:  config.openshift.io/v1
    Fields Type:  FieldsV1
    fieldsV1:
      f:spec:
        f:encryption:
          .:
          f:type:
        f:servingCerts:
          .:
          f:namedCertificates:
    Manager:         config-policy-controller
    Operation:       Update
    Time:            2021-07-27T09:23:32Z
  Resource Version:  61339482
  UID:               732a405d-9ea9-449e-8afe-9a58b977d947
Spec:
  Additional CORS Allowed Origins:
    (?i)//migration-openshift-migration\.apps\.eu-dev\.ocp\.aws\.boehringer\.com(:|\z)
    //127.0.0.1(:|$)
    //localhost(:|$)
  Audit:
    Profile:  Default
  Encryption:
    Type:  aescbc
  Serving Certs:
    Named Certificates:
      Names:
        api.eu-dev.ocp.aws.boehringer.com
      Serving Certificate:
        Name:  bi-api-certificate
Events:        <none>
[eu-dev@aws-us-bastion ~]$
~~~

[2] He confirms eu-dev/eu-dev-0-zpcgh-admin-kubeconfig is not working and have a query ***how can we update this to allow for our own certificates?***

~~~
[root@aws-us-bastion ~]# oc --kubeconfig=eu-dev.kubeconfig version
Client Version: 4.8.17
Unable to connect to the server: x509: certificate signed by unknown authority
[root@aws-us-bastion ~]#
~~~

Do let me know if more information is required from Cu's side.

Regards,
Utkarsh Wagh

Comment 23 daliu 2021-12-22 09:30:07 UTC
@jdiaz 
The customer has changed the certificates https://bugzilla.redhat.com/show_bug.cgi?id=2005759#c22,
Is there a detail doc about how to add the customer certificates in clusterdeployment? and for current customer, could you give a steps to make the clusterdeployment running? 
Thank you.

Comment 24 Joel Diaz 2021-12-22 13:46:19 UTC
There should be two options to get hive to trust the remote cluster's certificates.

1) Let hive apply the new certificates to the managed cluster:

This involves populating the ClusterDeployment's spec.certificateBundles[] list https://github.com/openshift/hive/blob/master/apis/hive/v1/clusterdeployment_types.go#L638-L653 . Give the certificate bundle containing the certs for the 'api' route a name, and create a Secret containing the certificate information (keep 'generate' set to 'false' as setting 'generate' to 'true' would have no effect as hive doesn't support generating certificates today).

Then you update the ClusterDeployment's spec.controlPlaneConfig.servingCertificates.default https://github.com/openshift/hive/blob/master/apis/hive/v1/clusterdeployment_types.go#L615-L619 to reference the name of the certificate bundle added under spec.certificateBundles[].

Hive will apply the certificates defined in the certificateBundle Secret for the 'api.<cluster>.domain.com' route, and will use the Certificate Authority information in that Secret when connecting to the cluster.

2) Tell hive that it should trust additional Certificate Authorities for all connections:

This involves configuring the HiveConfig .spec.additionalCertificateAuthoritiesSecretRef[] https://github.com/openshift/hive/blob/master/apis/hive/v1/hiveconfig_types.go#L29-L34 to tell hive about additional Certificate Authorities that hive should trust when making connections.

Comment 25 Utkarsh Wagh 2021-12-23 09:34:29 UTC
Hello Team,

I have an update from the CU side. They have followed [2] as mentioned above and have the following query:

~~~
We have decided to use [2] since all our clusters are similarly signed, we created the following secret and updated hiveconfig as documented but still not able to get it to work. Perhaps you can share more details on the exact steps to configure it? Thank you.
~~~

~~~
[us-acm@aws-us-bastion ~]$ oc describe secret -n hive bi-ca
Name:         bi-ca
Namespace:    hive
Labels:       <none>
Annotations:  <none>

Type:  Opaque

Data
====
bundle.cer:  5364 bytes
[us-acm@aws-us-bastion ~]$ oc describe hiveconfig
Name:         hive
Namespace:
Labels:       installer.name=multiclusterhub
              installer.namespace=open-cluster-management
Annotations:  <none>
API Version:  hive.openshift.io/v1
Kind:         HiveConfig
Metadata:
  Creation Timestamp:  2021-05-20T04:28:31Z
  Generation:          2
  Managed Fields:
    API Version:  hive.openshift.io/v1
    Fields Type:  FieldsV1
    fieldsV1:
      f:metadata:
        f:labels:
          .:
          f:installer.name:
          f:installer.namespace:
      f:spec:
        .:
        f:failedProvisionConfig:
          .:
          f:skipGatherLogs:
    Manager:      multiclusterhub-operator
    Operation:    Update
    Time:         2021-05-20T04:28:31Z
    API Version:  hive.openshift.io/v1
    Fields Type:  FieldsV1
    fieldsV1:
      f:spec:
        f:additionalCertificateAuthoritiesSecretRef:
    Manager:      kubectl-edit
    Operation:    Update
    Time:         2021-12-23T03:39:10Z
    API Version:  hive.openshift.io/v1
    Fields Type:  FieldsV1
    fieldsV1:
      f:spec:
        f:argoCDConfig:
          .:
          f:enabled:
        f:backup:
          .:
          f:velero:
        f:serviceProviderCredentialsConfig:
      f:status:
        f:aggregatorClientCAHash:
        f:conditions:
        f:configApplied:
        f:observedGeneration:
    Manager:         hive-operator
    Operation:       Update
    Time:            2021-12-23T03:39:11Z
  Resource Version:  648198222
  UID:               dc2ab216-f5f0-4e61-a493-8c5e4d70c1d4
Spec:
  Additional Certificate Authorities Secret Ref:
    Name:  bi-ca
  Failed Provision Config:
    Skip Gather Logs:  true
Status:
  Aggregator Client CA Hash:  38aee507acba7dd44dfb030e2d8711d9
  Conditions:
    Last Probe Time:       2021-09-22T03:50:30Z
    Last Transition Time:  2021-09-22T03:50:30Z
    Message:               Hive is deployed successfully
    Reason:                DeploymentSuccess
    Status:                True
    Type:                  Ready
  Config Applied:          true
  Observed Generation:     2
Events:                    <none>
[us-acm@aws-us-bastion ~]$
~~~

Do let me know if more data is required from the CU side.

Regards,
Utkarsh Wagh

Comment 26 Joel Diaz 2021-12-23 13:38:38 UTC
The certs in the Secret should be stored in the key named 'ca.crt' (not 'bundle.cer').

Comment 28 zyin@redhat.com 2022-01-05 10:37:43 UTC
have to clean up the resources in open-cluster-management-agent-addon project manually since the status of cluster is unknown.

1. check if there are failed jobs in open-cluster-management-agent-addon on the managed cluster. if yes, force delete them.
oc get jobs -n open-cluster-management-agent-addon
oc delete jobs xxx -n open-cluster-management-agent-addon

2. remove all finalizers of the deployments in the open-cluster-management-agent-addon project on the managed cluster manually.

3. force delete these appliedmanifestworks.work.open-cluster-management.io CRs.
oc delete appliedmanifestworks.work.open-cluster-management.io \

xxxx-<cluster name>-klusterlet-addon-appmgr         
xxxx-<cluster name>-klusterlet-addon-certpolicyctrl 
xxxx-<cluster name>-klusterlet-addon-iampolicyctrl        
xxxx-<cluster name>-klusterlet-addon-policyctrl     
xxxx-<cluster name>-klusterlet-addon-workmgr
xxxx-<cluster name>-klusterlet-addon-operator    

4. force delete the agent CRs. remove the finalizers manually if they have during the deletion.
oc delete applicationmanagers.agent.open-cluster-management.io klusterlet-addon-appmgr -n open-cluster-management-agent-addon
oc delete certpolicycontrollers.agent.open-cluster-management.io klusterlet-addon-certpolicyctrl -n open-cluster-management-agent-addon
oc delete iampolicycontrollers.agent.open-cluster-management.io klusterlet-addon-iampolicyctrl -n open-cluster-management-agent-addon
oc delete policycontrollers.agent.open-cluster-management.io klusterlet-addon-policyctrl -n open-cluster-management-agent-addon
oc delete searchcollectors.agent.open-cluster-management.io klusterlet-addon-search -n open-cluster-management-agent-addon
oc delete workmanagers.agent.open-cluster-management.io klusterlet-addon-workmgr -n open-cluster-management-agent-addon

Comment 29 Utkarsh Wagh 2022-01-06 04:57:49 UTC
Hello Team,

There is an update from the CU side for the above.

No failed jobs or deployments in open-cluster-management-agent-addon and appliedmanifestworks.work.open-cluster-management.io CRs, only the agent CRs exist on the managed clusters.


Please let us know how to force delete the agent CRs (and remove the finalizers manually if they have).

~~~
[root@aws-us-bastion ~]# om - get jobs -n open-cluster-management-agent-addon

[[ us-acm ]]
No resources found in open-cluster-management-agent-addon namespace.

[[ us-test ]]
No resources found in open-cluster-management-agent-addon namespace.

[[ us-dev ]]
No resources found in open-cluster-management-agent-addon namespace.

[[ us-prod ]]
No resources found in open-cluster-management-agent-addon namespace.

[[ eu-dev ]]
No resources found in open-cluster-management-agent-addon namespace.

[[ eu-prod ]]
No resources found in open-cluster-management-agent-addon namespace.

[root@aws-us-bastion ~]# om - get deployments -n open-cluster-management-agent-addon

[[ us-acm ]]
NAME                                        READY   UP-TO-DATE   AVAILABLE   AGE
klusterlet-addon-appmgr                     1/1     1            1           230d
klusterlet-addon-certpolicyctrl             1/1     1            1           230d
klusterlet-addon-iampolicyctrl              1/1     1            1           230d
klusterlet-addon-operator                   1/1     1            1           230d
klusterlet-addon-policyctrl-config-policy   1/1     1            1           230d
klusterlet-addon-policyctrl-framework       1/1     1            1           230d
klusterlet-addon-workmgr                    1/1     1            1           230d

[[ us-test ]]
No resources found in open-cluster-management-agent-addon namespace.

[[ us-dev ]]
No resources found in open-cluster-management-agent-addon namespace.

[[ us-prod ]]
No resources found in open-cluster-management-agent-addon namespace.

[[ eu-dev ]]
No resources found in open-cluster-management-agent-addon namespace.

[[ eu-prod ]]
No resources found in open-cluster-management-agent-addon namespace.

[root@aws-us-bastion ~]# om - get appliedmanifestworks.work.open-cluster-management.io -n open-cluster-management-agent-addon

[[ us-acm ]]
NAME                                                                                                             AGE
50afaaf31ceafe61ee70473433780c8f3b36a5046257ddab1dbfeb9ee1c5d3d9-local-cluster-klusterlet                        232d
50afaaf31ceafe61ee70473433780c8f3b36a5046257ddab1dbfeb9ee1c5d3d9-local-cluster-klusterlet-addon-appmgr           230d
50afaaf31ceafe61ee70473433780c8f3b36a5046257ddab1dbfeb9ee1c5d3d9-local-cluster-klusterlet-addon-certpolicyctrl   230d
50afaaf31ceafe61ee70473433780c8f3b36a5046257ddab1dbfeb9ee1c5d3d9-local-cluster-klusterlet-addon-crds             230d
50afaaf31ceafe61ee70473433780c8f3b36a5046257ddab1dbfeb9ee1c5d3d9-local-cluster-klusterlet-addon-iampolicyctrl    230d
50afaaf31ceafe61ee70473433780c8f3b36a5046257ddab1dbfeb9ee1c5d3d9-local-cluster-klusterlet-addon-operator         230d
50afaaf31ceafe61ee70473433780c8f3b36a5046257ddab1dbfeb9ee1c5d3d9-local-cluster-klusterlet-addon-policyctrl       230d
50afaaf31ceafe61ee70473433780c8f3b36a5046257ddab1dbfeb9ee1c5d3d9-local-cluster-klusterlet-addon-workmgr          230d
50afaaf31ceafe61ee70473433780c8f3b36a5046257ddab1dbfeb9ee1c5d3d9-local-cluster-klusterlet-crds                   230d
50afaaf31ceafe61ee70473433780c8f3b36a5046257ddab1dbfeb9ee1c5d3d9-local-cluster-observability                     91d

[[ us-test ]]
error: the server doesn't have a resource type "appliedmanifestworks"

[[ us-dev ]]
error: the server doesn't have a resource type "appliedmanifestworks"

[[ us-prod ]]
error: the server doesn't have a resource type "appliedmanifestworks"

[[ eu-dev ]]
error: the server doesn't have a resource type "appliedmanifestworks"

[[ eu-prod ]]
error: the server doesn't have a resource type "appliedmanifestworks"

[root@aws-us-bastion ~]# om - get applicationmanagers.agent.open-cluster-management.io klusterlet-addon-appmgr -n open-cluster-management-agent-addon

[[ us-acm ]]
NAME                      AGE
klusterlet-addon-appmgr   230d

[[ us-test ]]
NAME                      AGE
klusterlet-addon-appmgr   223d

[[ us-dev ]]
NAME                      AGE
klusterlet-addon-appmgr   211d

[[ us-prod ]]
NAME                      AGE
klusterlet-addon-appmgr   204d

[[ eu-dev ]]
NAME                      AGE
klusterlet-addon-appmgr   188d

[[ eu-prod ]]
NAME                      AGE
klusterlet-addon-appmgr   183d

[root@aws-us-bastion ~]# om - get certpolicycontrollers.agent.open-cluster-management.io klusterlet-addon-certpolicyctrl -n open-cluster-management-agent-addon

[[ us-acm ]]
NAME                              AGE
klusterlet-addon-certpolicyctrl   230d

[[ us-test ]]
NAME                              AGE
klusterlet-addon-certpolicyctrl   223d

[[ us-dev ]]
NAME                              AGE
klusterlet-addon-certpolicyctrl   211d

[[ us-prod ]]
NAME                              AGE
klusterlet-addon-certpolicyctrl   204d

[[ eu-dev ]]
NAME                              AGE
klusterlet-addon-certpolicyctrl   188d

[[ eu-prod ]]
NAME                              AGE
klusterlet-addon-certpolicyctrl   183d

[root@aws-us-bastion ~]# om - get iampolicycontrollers.agent.open-cluster-management.io klusterlet-addon-iampolicyctrl -n open-cluster-management-agent-addon

[[ us-acm ]]
NAME                             AGE
klusterlet-addon-iampolicyctrl   230d

[[ us-test ]]
NAME                             AGE
klusterlet-addon-iampolicyctrl   223d

[[ us-dev ]]
NAME                             AGE
klusterlet-addon-iampolicyctrl   211d

[[ us-prod ]]
NAME                             AGE
klusterlet-addon-iampolicyctrl   204d

[[ eu-dev ]]
NAME                             AGE
klusterlet-addon-iampolicyctrl   188d

[[ eu-prod ]]
NAME                             AGE
klusterlet-addon-iampolicyctrl   183d

[root@aws-us-bastion ~]# om - get policycontrollers.agent.open-cluster-management.io klusterlet-addon-policyctrl -n open-cluster-management-agent-addon

[[ us-acm ]]
NAME                          AGE
klusterlet-addon-policyctrl   230d

[[ us-test ]]
NAME                          AGE
klusterlet-addon-policyctrl   223d

[[ us-dev ]]
NAME                          AGE
klusterlet-addon-policyctrl   211d

[[ us-prod ]]
NAME                          AGE
klusterlet-addon-policyctrl   204d

[[ eu-dev ]]
NAME                          AGE
klusterlet-addon-policyctrl   188d

[[ eu-prod ]]
NAME                          AGE
klusterlet-addon-policyctrl   183d

[root@aws-us-bastion ~]# om - get searchcollectors.agent.open-cluster-management.io klusterlet-addon-search -n open-cluster-management-agent-addon

[[ us-acm ]]
Error from server (NotFound): searchcollectors.agent.open-cluster-management.io "klusterlet-addon-search" not found

[[ us-test ]]
NAME                      AGE
klusterlet-addon-search   223d

[[ us-dev ]]
NAME                      AGE
klusterlet-addon-search   211d

[[ us-prod ]]
NAME                      AGE
klusterlet-addon-search   204d

[[ eu-dev ]]
NAME                      AGE
klusterlet-addon-search   188d

[[ eu-prod ]]
NAME                      AGE
klusterlet-addon-search   183d

[root@aws-us-bastion ~]# om - get workmanagers.agent.open-cluster-management.io klusterlet-addon-workmgr -n open-cluster-management-agent-addon

[[ us-acm ]]
om - get workmanagers.agent.open-cluster-management.io klusterlet-addon-workmgr -n open-cluster-management-agent-addon
NAME                       AGE
klusterlet-addon-workmgr   230d

[[ us-test ]]
NAME                       AGE
klusterlet-addon-workmgr   223d

[[ us-dev ]]
NAME                       AGE
klusterlet-addon-workmgr   211d

[[ us-prod ]]
NAME                       AGE
klusterlet-addon-workmgr   204d

[[ eu-dev ]]
NAME                       AGE
klusterlet-addon-workmgr   188d

[[ eu-prod ]]
NAME                       AGE
klusterlet-addon-workmgr   183d

[root@aws-us-bastion ~]#
~~~

Let me know if more information is required from CU's side.

Regards,

Comment 30 zyin@redhat.com 2022-01-06 06:10:54 UTC
it looks like the work agents are not working except the one on the hub cluster.


login to the managed cluster, delete the agent CRs on the managed cluster (not on the hub cluster)

oc delete applicationmanagers.agent.open-cluster-management.io klusterlet-addon-appmgr -n open-cluster-management-agent-addon
oc delete certpolicycontrollers.agent.open-cluster-management.io klusterlet-addon-certpolicyctrl -n open-cluster-management-agent-addon
oc delete iampolicycontrollers.agent.open-cluster-management.io klusterlet-addon-iampolicyctrl -n open-cluster-management-agent-addon
oc delete policycontrollers.agent.open-cluster-management.io klusterlet-addon-policyctrl -n open-cluster-management-agent-addon
oc delete searchcollectors.agent.open-cluster-management.io klusterlet-addon-search -n open-cluster-management-agent-addon
oc delete workmanagers.agent.open-cluster-management.io klusterlet-addon-workmgr -n open-cluster-management-agent-addon

if there are some other resources which prevent the project open-cluster-management-agent-addon from being deleted, delete them manually too.

Comment 31 Yuanyuan He 2022-01-12 06:33:51 UTC
Can the steps appended above helps to force delete the agent CRs, and is there still any problems? Thanks!

Comment 32 Utkarsh Wagh 2022-01-12 06:38:47 UTC
Hello Team,

~~~
Can the steps append above helps to force delete the agent CRs, and are there still any problems? Thanks!
~~~

Yes, the CU has updated that the issue is resolved and I have archived the case. Thank you for your help.


Regards,
Utkarsh Wagh

Comment 33 Yuanyuan He 2022-01-12 08:03:56 UTC
Glad to know, I'll close the issue then and feel free to reopen in case of any further problems. Thank you!


Note You need to log in before you can comment on or make changes to this bug.